Predict 5' and 3' UTR's using RNA-seq?

[James]

Hi all,

I looked quickly but I couldn't see any threads.

The organism I work on has poorly annotated Untranslated regions (UTRs) There is some EST data I could use, but really I was planning on using our ever increasing RNA-seq data to do this.

Does anybody know any tools that can do this? or any scripts people may want to give me? or good ideas on how to go about this.

Thanks, J

[steven]

Are there a lot of introns in that genome? If not, ORF detection in each read cluster could be an option, but it will depend on the coverage and on the proportion of repeats.
If you already have ORFs but no good UTRs (like in yeast I believe) you may as well just extend them by projecting/collapsing/merging the reads on each side to define the UTRs. Something like BEDtools could work for that. I am assuming you know the genomic sequence of that organism of course..
I hope it makes sense.

[frozenlyse]

You could try CAGE-seq to annotate 5'-ends of RNAs, that combined with RNA-seq will vastly help annotate 5' UTR's at least.